Coverage for session_buddy/utils/regex

1#!/usr/bin/env python3

2"""Local regex patterns for session-mgmt-mcp using crackerjack's SAFE_PATTERNS architecture.

4This module contains all validated regex patterns used throughout the session-mgmt-mcp

5codebase, following crackerjack's centralized pattern approach for security and reliability.

7All patterns are validated with comprehensive test cases and use proper replacement syntax.

8"""

10import re

12from crackerjack.services.regex_patterns import ValidatedPattern

14# Session Management MCP Validated Patterns Registry

15SAFE_PATTERNS: dict[str, ValidatedPattern] = {

16 # Code extraction patterns for search_enhanced.py

17 "python_code_block": ValidatedPattern(

18 name="python_code_block",

19 pattern=r"```python\n(.*?)\n```",

20 replacement=r"\1",

21 description="Extract Python code from markdown code blocks",

22 flags=re.DOTALL,

23 test_cases=[

24 ("```python\nprint('hello')\n```", "print('hello')"),

25 ("```python\ndef func():\n pass\n```", "def func():\n pass"),

26 ("```python\n\n```", ""),

27 ("no code here", "no code here"), # No change

28 ],

29 ),

30 "generic_code_block": ValidatedPattern(

31 name="generic_code_block",

32 pattern=r"```\n(.*?)\n```",

33 replacement=r"\1",

34 description="Extract code from generic markdown code blocks",

35 flags=re.DOTALL,

36 test_cases=[

37 ("```\nsome code\n```", "some code"),

38 ("```\nline1\nline2\n```", "line1\nline2"),

39 ("```\n\n```", ""),

40 ("no fenced code", "no fenced code"), # No change

41 ],

42 ),

43 # Error pattern matching for search_enhanced.py

44 "python_traceback": ValidatedPattern(

45 name="python_traceback",

46 pattern=r"Traceback $most recent call last$:.*?(?=\n\n|\Z)",

47 replacement=r"<TRACEBACK_MASKED>",

48 description="Match Python traceback blocks with safe termination",

49 flags=re.MULTILINE | re.DOTALL,

50 test_cases=[

51 (

52 "Traceback (most recent call last):\n File test.py\nError: msg\n\nNext line",

53 "<TRACEBACK_MASKED>\n\nNext line",

54 ),

55 (

56 "Traceback (most recent call last):\n File test.py\nError: msg",

57 "<TRACEBACK_MASKED>",

58 ),

59 ("No traceback here", "No traceback here"), # No change

60 ],

61 ),

62 "python_exception": ValidatedPattern(

63 name="python_exception",

65 replacement=r"\1: <ERROR_MESSAGE_MASKED>",

66 description="Match Python exception patterns safely",

67 test_cases=[

68 ("ValueError: invalid input", "ValueError: <ERROR_MESSAGE_MASKED>"),

69 (

70 "RuntimeError: something went wrong",

71 "RuntimeError: <ERROR_MESSAGE_MASKED>",

72 ),

73 (

74 "NotAnError: this should not match",

75 "NotAnError: this should not match",

76 ), # No change

77 ("SyntaxError: bad syntax", "SyntaxError: <ERROR_MESSAGE_MASKED>"),

78 ],

79 ),

80 "javascript_error": ValidatedPattern(

81 name="javascript_error",

82 pattern=r"\b(Error|TypeError|ReferenceError): (.+)",

83 replacement=r"\1: <JS_ERROR_MASKED>",

84 description="Match JavaScript error patterns",

85 test_cases=[

86 ("TypeError: Cannot read property", "TypeError: <JS_ERROR_MASKED>"),

87 ("Error: Something failed", "Error: <JS_ERROR_MASKED>"),

88 ("ReferenceError: x is not defined", "ReferenceError: <JS_ERROR_MASKED>"),

89 ("CustomError: not matched", "CustomError: not matched"), # No change

90 ],

91 ),

92 "compile_error": ValidatedPattern(

93 name="compile_error",

94 pattern=r"(error|Error): (.+) at line (\d+)",

95 replacement=r"\1: <COMPILE_ERROR_MASKED> at line \3",

96 description="Match compilation error patterns with line numbers",

97 test_cases=[

98 (

99 "error: syntax error at line 42",

100 "error: <COMPILE_ERROR_MASKED> at line 42",

101 ),

102 (

103 "Error: missing semicolon at line 10",

104 "Error: <COMPILE_ERROR_MASKED> at line 10",

105 ),

106 (

107 "warning: deprecated at line 5",

108 "warning: deprecated at line 5",

109 ), # No change

110 ],

111 ),

112 "warning_pattern": ValidatedPattern(

113 name="warning_pattern",

114 pattern=r"(warning|Warning): (.+)",

115 replacement=r"\1: <WARNING_MASKED>",

116 description="Match warning message patterns",

117 test_cases=[

118 ("warning: deprecated function", "warning: <WARNING_MASKED>"),

119 ("Warning: potential issue", "Warning: <WARNING_MASKED>"),

120 ("info: just information", "info: just information"), # No change

121 ],

122 ),

123 "assertion_error": ValidatedPattern(

124 name="assertion_error",

125 pattern=r"AssertionError: (.+)",

126 replacement=r"AssertionError: <ASSERTION_MASKED>",

127 description="Match assertion error patterns",

128 test_cases=[

129 ("AssertionError: expected True", "AssertionError: <ASSERTION_MASKED>"),

130 (

131 "AssertionError: values don't match",

132 "AssertionError: <ASSERTION_MASKED>",

133 ),

134 ("ValueError: not assertion", "ValueError: not assertion"), # No change

135 ],

136 ),

137 "import_error": ValidatedPattern(

138 name="import_error",

139 pattern=r"ImportError: (.+)",

140 replacement=r"ImportError: <IMPORT_ERROR_MASKED>",

141 description="Match import error patterns",

142 test_cases=[

143 (

144 "ImportError: No module named 'xyz'",

145 "ImportError: <IMPORT_ERROR_MASKED>",

146 ),

147 ("ImportError: cannot import name", "ImportError: <IMPORT_ERROR_MASKED>"),

148 (

149 "ModuleNotFoundError: different",

150 "ModuleNotFoundError: different",

151 ), # No change

152 ],

153 ),

154 "module_not_found": ValidatedPattern(

155 name="module_not_found",

156 pattern=r"ModuleNotFoundError: (.+)",

157 replacement=r"ModuleNotFoundError: <MODULE_NOT_FOUND_MASKED>",

158 description="Match module not found error patterns",

159 test_cases=[

160 (

161 "ModuleNotFoundError: No module named 'test'",

162 "ModuleNotFoundError: <MODULE_NOT_FOUND_MASKED>",

163 ),

164 (

165 "ModuleNotFoundError: missing dependency",

166 "ModuleNotFoundError: <MODULE_NOT_FOUND_MASKED>",

167 ),

168 (

169 "ImportError: different error",

170 "ImportError: different error",

171 ), # No change

172 ],

173 ),

174 "file_not_found": ValidatedPattern(

175 name="file_not_found",

176 pattern=r"FileNotFoundError: (.+)",

177 replacement=r"FileNotFoundError: <FILE_NOT_FOUND_MASKED>",

178 description="Match file not found error patterns",

179 test_cases=[

180 (

181 "FileNotFoundError: [Errno 2] No such file",

182 "FileNotFoundError: <FILE_NOT_FOUND_MASKED>",

183 ),

184 (

185 "FileNotFoundError: file missing",

186 "FileNotFoundError: <FILE_NOT_FOUND_MASKED>",

187 ),

188 ("PermissionError: different", "PermissionError: different"), # No change

189 ],

190 ),

191 "permission_denied": ValidatedPattern(

192 name="permission_denied",

193 pattern=r"PermissionError: (.+)",

194 replacement=r"PermissionError: <PERMISSION_ERROR_MASKED>",

195 description="Match permission error patterns",

196 test_cases=[

197 (

198 "PermissionError: [Errno 13] Permission denied",

199 "PermissionError: <PERMISSION_ERROR_MASKED>",

200 ),

201 (

202 "PermissionError: access denied",

203 "PermissionError: <PERMISSION_ERROR_MASKED>",

204 ),

205 (

206 "FileNotFoundError: different",

207 "FileNotFoundError: different",

208 ), # No change

209 ],

210 ),

211 "network_error": ValidatedPattern(

212 name="network_error",

213 pattern=r"(ConnectionError|TimeoutError|HTTPError): (.+)",

214 replacement=r"\1: <NETWORK_ERROR_MASKED>",

215 description="Match network-related error patterns",

216 test_cases=[

217 (

218 "ConnectionError: Failed to connect",

219 "ConnectionError: <NETWORK_ERROR_MASKED>",

220 ),

221 ("TimeoutError: Request timed out", "TimeoutError: <NETWORK_ERROR_MASKED>"),

222 ("HTTPError: 404 Not Found", "HTTPError: <NETWORK_ERROR_MASKED>"),

223 ("ValueError: not network", "ValueError: not network"), # No change

224 ],

225 ),

226 # Context pattern matching (boolean search)

227 "debugging_context": ValidatedPattern(

228 name="debugging_context",

229 pattern=r"\b(debug|debugging|breakpoint|pdb|print)\b",

230 replacement=r"<DEBUG_CONTEXT>",

231 description="Match debugging-related context patterns",

232 flags=re.IGNORECASE,

233 test_cases=[

234 ("let's debug this issue", "let's <DEBUG_CONTEXT> this issue"),

235 ("debugging the problem", "<DEBUG_CONTEXT> the problem"),

236 ("set a breakpoint here", "set a <DEBUG_CONTEXT> here"),

237 ("regular code", "regular code"), # No change

238 ],

239 ),

240 "testing_context": ValidatedPattern(

241 name="testing_context",

242 pattern=r"(test|pytest|unittest|assert|mock)",

243 replacement=r"<TEST_CONTEXT>",

244 description="Match testing-related context patterns",

245 flags=re.IGNORECASE,

246 test_cases=[

247 ("run the test suite", "run the <TEST_CONTEXT> suite"),

248 ("pytest configuration", "<TEST_CONTEXT> configuration"),

249 ("unittest framework", "<TEST_CONTEXT> framework"),

250 ("regular text", "regular text"), # No change

251 ],

252 ),

253 "error_handling_context": ValidatedPattern(

254 name="error_handling_context",

255 pattern=r"(try|except|finally|raise|catch)",

256 replacement=r"<ERROR_HANDLING_CONTEXT>",

257 description="Match error handling context patterns",

258 flags=re.IGNORECASE,

259 test_cases=[

260 ("try to handle", "<ERROR_HANDLING_CONTEXT> to handle"),

261 ("except ValueError", "<ERROR_HANDLING_CONTEXT> ValueError"),

262 ("finally block", "<ERROR_HANDLING_CONTEXT> block"),

263 ("normal flow", "normal flow"), # No change

264 ],

265 ),

266 "performance_context": ValidatedPattern(

267 name="performance_context",

268 pattern=r"(slow|performance|benchmark|optimize|profil)",

269 replacement=r"<PERFORMANCE_CONTEXT>",

270 description="Match performance-related context patterns",

271 flags=re.IGNORECASE,

272 test_cases=[

273 ("this is slow", "this is <PERFORMANCE_CONTEXT>"),

274 ("performance improvement", "<PERFORMANCE_CONTEXT> improvement"),

275 ("benchmark results", "<PERFORMANCE_CONTEXT> results"),

276 ("fast code", "fast code"), # No change

277 ],

278 ),

279 "security_context": ValidatedPattern(

280 name="security_context",

281 pattern=r"(security|authentication|authorization|token|password)",

282 replacement=r"<SECURITY_CONTEXT>",

283 description="Match security-related context patterns",

284 flags=re.IGNORECASE,

285 test_cases=[

286 ("security audit", "<SECURITY_CONTEXT> audit"),

287 ("authentication required", "<SECURITY_CONTEXT> required"),

288 ("token validation", "<SECURITY_CONTEXT> validation"),

289 ("regular text", "regular text"), # No change

290 ],

291 ),

292 # Time parsing patterns for search_enhanced.py

293 "time_ago_pattern": ValidatedPattern(

294 name="time_ago_pattern",

296 replacement=r"\1 \2 ago",

297 description="Match time ago expressions for parsing",

298 test_cases=[

299 ("5 minutes ago", "5 minute ago"),

300 ("2 hours ago", "2 hour ago"),

301 ("1 day ago", "1 day ago"),

302 ("3 weeks ago", "3 week ago"),

303 ("not a time", "not a time"), # No change

304 ],

305 ),

306 "relative_time_pattern": ValidatedPattern(

307 name="relative_time_pattern",

309 replacement=r"<RELATIVE_TIME>",

310 description="Match relative time expressions",

311 flags=re.IGNORECASE,

312 test_cases=[

313 ("today is good", "<RELATIVE_TIME> is good"),

314 ("yesterday we worked", "<RELATIVE_TIME> we worked"),

315 ("this week's plan", "<RELATIVE_TIME>'s plan"),

316 ("some other day", "some other day"), # No change

317 ],

318 ),

319 "since_time_pattern": ValidatedPattern(

320 name="since_time_pattern",

321 pattern=r"since\s+(today|yesterday|this\s+week|last\s+week)",

322 replacement=r"since <TIME_REFERENCE>",

323 description="Match 'since' time expressions",

324 flags=re.IGNORECASE,

325 test_cases=[

326 ("since today", "since <TIME_REFERENCE>"),

327 ("since yesterday", "since <TIME_REFERENCE>"),

328 ("since this week", "since <TIME_REFERENCE>"),

329 ("since forever", "since forever"), # No change

330 ],

331 ),

332 "last_duration_pattern": ValidatedPattern(

333 name="last_duration_pattern",

335 replacement=r"in the last \1 \2",

336 description="Match 'in the last X units' patterns",

337 test_cases=[

338 ("in the last 5 minutes", "in the last 5 minute"),

339 ("in the last 2 hours", "in the last 2 hour"),

340 ("in the last 10 days", "in the last 10 day"),

341 ("not a duration", "not a duration"), # No change

342 ],

343 ),

344 "iso_date_pattern": ValidatedPattern(

345 name="iso_date_pattern",

346 pattern=r"(\d{4}-(0[1-9]|1[0-2])-(0[1-9]|[12]\d|3[01]))",

347 replacement=r"<ISO_DATE>",

348 description="Match valid ISO date format (YYYY-MM-DD)",

349 test_cases=[

350 ("2023-12-25", "<ISO_DATE>"),

351 ("Date: 2024-01-15 is today", "Date: <ISO_DATE> is today"),

352 ("not-a-date", "not-a-date"), # No change

353 (

354 "2023-13-45",

355 "2023-13-45",

356 ), # Invalid date, no change expected

357 ],

358 ),

359 "us_date_pattern": ValidatedPattern(

360 name="us_date_pattern",

361 pattern=r"(\d{1,2}/\d{1,2}/\d{4})",

362 replacement=r"<US_DATE>",

363 description="Match US date format (MM/DD/YYYY or M/D/YYYY)",

364 test_cases=[

365 ("12/25/2023", "<US_DATE>"),

366 ("1/5/2024", "<US_DATE>"),

367 ("Meeting on 3/15/2024 at noon", "Meeting on <US_DATE> at noon"),

368 ("not/a/date", "not/a/date"), # No change

369 ],

370 ),

371 # Crackerjack integration patterns for output parsing

372 "pytest_result": ValidatedPattern(

373 name="pytest_result",

375 replacement=r"TEST: \1::\2 -> \3",

376 description="Parse pytest test results with optional percentage and timing",

377 test_cases=[

378 (

379 "test_file.py:: test_function PASSED",

380 "TEST: test_file.py::test_function -> PASSED",

381 ),

382 (

383 "test_example.py:: test_method FAILED [50%] (0.05s)",

384 "TEST: test_example.py::test_method -> FAILED",

385 ),

386 (

387 "test_skip.py:: test_skip SKIPPED",

388 "TEST: test_skip.py::test_skip -> SKIPPED",

389 ),

390 ("not a test result", "not a test result"), # No change

391 ],

392 ),

393 "coverage_summary": ValidatedPattern(

394 name="coverage_summary",

395 pattern=r"TOTAL\s+\d+\s+\d+\s+(\d+)%",

396 replacement=r"COVERAGE: \1%",

397 description="Extract total coverage percentage from coverage reports",

398 test_cases=[

399 ("TOTAL 1000 50 95%", "COVERAGE: 95%"),

400 ("TOTAL 500 25 78%", "COVERAGE: 78%"),

401 ("subtotal 100 5 90%", "subtotal 100 5 90%"), # No change

402 ],

403 ),

404 "ruff_error": ValidatedPattern(

405 name="ruff_error",

406 pattern=r"([^:\s]+):(\d+):(\d+):\s*([A-Z]\d{3,4})\s*(.+)",

407 replacement=r"RUFF: \1 line \2 -> \4: \5",

408 description="Parse Ruff linting errors with file, line, column, code, and message",

409 test_cases=[

410 (

411 "src/main.py:42:10: E501 line too long (88 > 79 characters)",

412 "RUFF: src/main.py line 42 -> E501: line too long (88 > 79 characters)",

413 ),

414 (

415 "test.py:1:1: F401 imported but unused",

416 "RUFF: test.py line 1 -> F401: imported but unused",

417 ),

418 ("not a ruff error", "not a ruff error"), # No change

419 ],

420 ),

421 "mypy_error": ValidatedPattern(

422 name="mypy_error",

423 pattern=r"([^:\s]+):(\d+):\s*error:\s*(.+)",

424 replacement=r"MYPY: \1 line \2 -> \3",

425 description="Parse mypy type checking errors with file, line, and message",

426 test_cases=[

427 (

428 "src/module.py:15: error: Argument 1 has incompatible type",

429 "MYPY: src/module.py line 15 -> Argument 1 has incompatible type",

430 ),

431 (

432 "main.py:8: error: Name 'x' is not defined",

433 "MYPY: main.py line 8 -> Name 'x' is not defined",

434 ),

435 ("not a mypy error", "not a mypy error"), # No change

436 ],

437 ),

438 "bandit_finding": ValidatedPattern(

439 name="bandit_finding",

440 pattern=r">> Issue: \[([A-Z]\d+):([a-z_]+)\]\s*(.+)",

441 replacement=r"BANDIT: \1 (\2) -> \3",

442 description="Parse Bandit security findings with code, severity, and description",

443 test_cases=[

444 (

445 ">> Issue: [B602:subprocess_popen_with_shell_equals_true] Possible shell injection",

446 "BANDIT: B602 (subprocess_popen_with_shell_equals_true) -> Possible shell injection",

447 ),

448 (

449 ">> Issue: [B108:hardcoded_tmp_directory] Use of insecure temp",

450 "BANDIT: B108 (hardcoded_tmp_directory) -> Use of insecure temp",

451 ),

452 ("not a bandit finding", "not a bandit finding"), # No change

453 ],

454 ),

455 "quality_score": ValidatedPattern(

456 name="quality_score",

457 pattern=r"Quality Score:\s*(\d+(?:\.\d+)?)/(\d+(?:\.\d+)?)\s*$(\d+(?:\.\d+)?)%$",

458 replacement=r"QUALITY: \3% (\1/\2)",

459 description="Parse quality score with percentage calculation",

460 test_cases=[

461 ("Quality Score: 85.5/100 (85.5%)", "QUALITY: 85.5% (85.5/100)"),

462 ("Quality Score: 90/100 (90%)", "QUALITY: 90% (90/100)"),

463 ("Final Score: 95/100", "Final Score: 95/100"), # No change

464 ],

465 ),

466 "execution_time": ValidatedPattern(

467 name="execution_time",

468 pattern=r"(\d+(?:\.\d+)?)\s*(s|ms|seconds?|milliseconds?)",

469 replacement=r"\1\2",

470 description="Normalize execution time formats",

471 test_cases=[

472 ("2.5 seconds", "2.5seconds"),

473 ("150 ms", "150ms"),

474 ("0.05 s", "0.05s"),

475 ("fast enough", "fast enough"), # No change

476 ],

477 ),

478 "progress_indicator": ValidatedPattern(

479 name="progress_indicator",

480 pattern=r"\[([=>\s]*)\]\s*(\d+)%",

481 replacement=r"PROGRESS: \2%",

482 description="Parse progress bars with percentage",

483 test_cases=[

484 ("[=====> ] 50%", "PROGRESS: 50%"),

485 ("[=========>] 90%", "PROGRESS: 90%"),

486 ("[ ] 0%", "PROGRESS: 0%"),

487 ("not progress", "not progress"), # No change

488 ],

489 ),

490 "git_commit_hash": ValidatedPattern(

491 name="git_commit_hash",

492 pattern=r"\b([a-f0-9]{7,40})\b",

493 replacement=r"<COMMIT:\1>",

494 description="Match Git commit hashes (7-40 hex characters)",

495 test_cases=[

496 ("commit abc1234 was merged", "commit <COMMIT:abc1234> was merged"),

497 (

498 "long hash abcdef1234567890abcdef1234567890abcdef12",

499 "long hash <COMMIT:abcdef1234567890abcdef1234567890abcdef12>",

500 ),

501 ("short ab12", "short ab12"), # Too short

502 ("not hex ghi1234", "not hex ghi1234"), # Contains non-hex

503 ],

504 ),

505 "file_path_with_line": ValidatedPattern(

506 name="file_path_with_line",

507 pattern=r"([A-Za-z_][A-Za-z0-9_/.-]*\.py):(\d+)",

508 replacement=r"\1 line \2",

509 description="Match file paths with line numbers (file.py:123)",

510 test_cases=[

511 ("src/main.py:42", "src/main.py line 42"),

512 ("test_file.py:1", "test_file.py line 1"),

513 ("sub/dir/module.py:100", "sub/dir/module.py line 100"),

514 ("not-python.txt:50", "not-python.txt:50"), # No change

515 ],

516 ),

517 # Memory optimizer patterns

518 "sentence_split": ValidatedPattern(

519 name="sentence_split",

520 pattern=r"[.!\?]+",

521 replacement=r" ",

522 description="Replace sentence-ending punctuation with spaces for splitting",

523 test_cases=[

524 ("Hello world.", "Hello world "),

525 ("How are you?", "How are you "),

526 ("Multiple!!!", "Multiple "),

527 ],

528 ),

529 "code_block_cleanup": ValidatedPattern(

530 name="code_block_cleanup",

531 pattern=r"```.*?```",

532 replacement=r"",

533 description="Remove code blocks from text for keyword extraction",

534 flags=re.DOTALL,

535 test_cases=[

536 ("Text ```python\ncode\n``` more text", "Text more text"),

537 ("```\njust code\n```", ""),

538 ("no code blocks", "no code blocks"),

539 ],

540 ),

541 "inline_code_cleanup": ValidatedPattern(

542 name="inline_code_cleanup",

543 pattern=r"`[^`]+`",

544 replacement=r"",

545 description="Remove inline code from text",

546 test_cases=[

547 ("Use `function()` to call", "Use to call"),

548 ("No inline code", "No inline code"),

549 ("`code`", ""),

550 ],

551 ),

552 "word_extraction": ValidatedPattern(

553 name="word_extraction",

554 pattern=r"\b[a-zA-Z]{1,2}\b",

555 replacement=r"",

556 description="Remove short words (1-2 chars) for keyword analysis",

557 test_cases=[

558 ("short words", "short words"),

559 ("a test", " test"),

560 ("123 in 456", "123 456"),

561 ],

562 ),

563 "word_boundary": ValidatedPattern(

564 name="word_boundary",

565 pattern=r"[^\w\s]+",

566 replacement=r" ",

567 description="Replace non-word characters with spaces for word boundary detection",

568 test_cases=[

569 ("hello-world", "hello world"),

570 ("test@example", "test example"),

571 ("underscored_var", "underscored_var"),

572 ],

573 ),

574 # File extension patterns for memory optimizer

575 "python_files": ValidatedPattern(

576 name="python_files",

577 pattern=r"(\w+\.py)",

578 replacement=r"[\1]",

579 description="Wrap Python file references in brackets",

580 test_cases=[

581 ("main.py script", "[main.py] script"),

582 ("test_file.py found", "[test_file.py] found"),

583 ("no files here", "no files here"),

584 ],

585 ),

586 "javascript_files": ValidatedPattern(

587 name="javascript_files",

588 pattern=r"(\w+\.js)",

589 replacement=r"[\1]",

590 description="Wrap JavaScript file references in brackets",

591 test_cases=[

592 ("app.js file", "[app.js] file"),

593 ("script.js found", "[script.js] found"),

594 ("no files", "no files"),

595 ],

596 ),

597 "typescript_files": ValidatedPattern(

598 name="typescript_files",

599 pattern=r"(\w+\.ts)",

600 replacement=r"[\1]",

601 description="Wrap TypeScript file references in brackets",

602 test_cases=[

603 ("index.ts file", "[index.ts] file"),

604 ("component.ts interface", "[component.ts] interface"),

605 ("other files", "other files"),

606 ],

607 ),

608 "json_files": ValidatedPattern(

609 name="json_files",

610 pattern=r"(\w+\.json)",

611 replacement=r"[\1]",

612 description="Match JSON file references",

613 test_cases=[

614 ("config.json settings", "[config.json] settings"),

615 ("package.json file", "[package.json] file"),

616 ("no json", "no json"),

617 ],

618 ),

619 "markdown_files": ValidatedPattern(

620 name="markdown_files",

621 pattern=r"(\w+\.md)",

622 replacement=r"[\1]",

623 description="Match Markdown file references",

624 test_cases=[

625 ("README.md documentation", "[README.md] documentation"),

626 ("docs.md file", "[docs.md] file"),

627 ("no markdown", "no markdown"),

628 ],

629 ),

630 # Advanced search patterns

631 "function_definition": ValidatedPattern(

632 name="function_definition",

633 pattern=r"\bdef\s+(\w+)",

634 replacement=r"function:\1",

635 description="Extract Python function definitions",

636 test_cases=[

637 ("def main():", "function:main():"),

638 ("def get_data(param):", "function:get_data(param):"),

639 ("no functions here", "no functions here"),

640 ],

641 ),

642 "class_definition": ValidatedPattern(

643 name="class_definition",

644 pattern=r"\bclass\s+(\w+)",

645 replacement=r"class:\1",

646 description="Extract Python class definitions",

647 test_cases=[

648 ("class MyClass:", "class:MyClass:"),

649 ("class SearchEngine(Base):", "class:SearchEngine(Base):"),

650 ("no classes here", "no classes here"),

651 ],

652 ),

653 "file_extension": ValidatedPattern(

654 name="file_extension",

655 pattern=r"\.(\w{2,4})\b",

656 replacement=r"filetype:\1",

657 description="Extract file extensions for categorization",

658 test_cases=[

659 ("file.py and test.json", "filefiletype:py and test.json"),

660 ("config.yaml setup", "configfiletype:yaml setup"),

661 ("no extensions", "no extensions"),

662 ],

663 ),

664 # Language detection patterns

665 "python_code": ValidatedPattern(

666 name="python_code",

668 replacement=r"python",

669 description="Detect Python code patterns",

670 flags=re.IGNORECASE,

671 test_cases=[

672 ("def function():", "python function():"),

673 ("import os", "python os"),

674 ("if __name__ == '__main__'", "python == '__main__'"),

675 ("regular text", "regular text"),

676 ],

677 ),

678 "javascript_code": ValidatedPattern(

679 name="javascript_code",

681 replacement=r"javascript",

682 description="Detect JavaScript code patterns",

683 flags=re.IGNORECASE,

684 test_cases=[

685 ("function myFunc()", "javascript myFunc()"),

686 ("const data = []", "javascript data = []"),

687 ("console.log('hello')", "javascript('hello')"),

688 ("regular text", "regular text"),

689 ],

690 ),

691 "sql_code": ValidatedPattern(

692 name="sql_code",

694 replacement=r"sql",

695 description="Detect SQL code patterns",

696 flags=re.IGNORECASE,

697 test_cases=[

698 ("SELECT * FROM users", "sql * FROM users"),

699 ("INSERT INTO table", "sql INTO table"),

700 ("CREATE TABLE test", "sql TABLE test"),

701 ("regular text", "regular text"),

702 ],

703 ),

704 "error_keywords": ValidatedPattern(

705 name="error_keywords",

707 replacement=r"error",

708 description="Detect error-related keywords",

709 flags=re.IGNORECASE,

710 test_cases=[

711 ("ValueError occurred", "error occurred"),

712 ("Exception raised", "error raised"),

713 ("Traceback found", "error found"),

714 ("regular text", "regular text"),

715 ],

716 ),

717 # Crackerjack-specific patterns for tools integration

718 "crackerjack_command": ValidatedPattern(

719 name="crackerjack_command",

720 pattern=r"crackerjack\s+(\w+)",

721 replacement=r"[\1]",

722 description="Extract command from crackerjack execution logs",

723 flags=0,

724 test_cases=[

725 ("crackerjack lint", "[lint]"),

726 ("running crackerjack test now", "running [test] now"),

727 ("crackerjack analyze completed", "[analyze] completed"),

728 ("just crackerjack", "just crackerjack"), # No change - no command

729 ],

730 ),

731 # Git and system security patterns

732 "safe_branch_name": ValidatedPattern(

733 name="safe_branch_name",

734 pattern=r"^[a-zA-Z0-9_/-]+$",

735 replacement=r"\g<0>",

736 description="Validate safe Git branch names (alphanumeric, dashes, underscores, slashes)",

737 flags=0,

738 test_cases=[

739 ("main", "main"),

740 ("feature/new-feature", "feature/new-feature"),

741 ("dev_branch", "dev_branch"),

742 ("hotfix-123", "hotfix-123"),

743 ("origin/main", "origin/main"),

744 ("invalid;branch", "invalid;branch"), # No match expected for invalid chars

745 ("branch with spaces", "branch with spaces"), # No match expected

746 ],

747 ),

748 # Token optimization patterns

749 "whitespace_normalize": ValidatedPattern(

750 name="whitespace_normalize",

751 pattern=r"\s+",

752 replacement=r" ",

753 description="Normalize whitespace for content hashing",

754 flags=0,

755 global_replace=True,

756 test_cases=[

757 (" multiple spaces ", " multiple spaces "),

758 ("tabs\t\tand\nnewlines\n\n", "tabs and newlines "),

759 ("normal text", "normal text"),

760 ],

761 ),

762}

Coverage for session_buddy / utils / regex_patterns.py: 100.00%

2 statements